Electrocardiogram Generation and Feature Extraction Using a Variational Autoencoder

TL;DR

This paper introduces a variational autoencoder-based method for generating realistic ECG signals and extracting interpretable features, enhancing diagnostic accuracy and addressing data scarcity in supervised learning.

Contribution

It presents a novel approach combining ECG generation and feature extraction with a variational autoencoder, improving data quality and interpretability for cardiovascular diagnostics.

Findings

Generated ECGs have high realism, indicated by a low MMD of 0.00383.

Extracted features are interpretable and can improve diagnostic models.

Synthetic ECGs can mitigate data scarcity for supervised learning.

Abstract

We propose a method for generating an electrocardiogram (ECG) signal for one cardiac cycle using a variational autoencoder. Using this method we extracted a vector of new 25 features, which in many cases can be interpreted. The generated ECG has quite natural appearance. The low value of the Maximum Mean Discrepancy metric, 0.00383, indicates good quality of ECG generation too. The extracted new features will help to improve the quality of automatic diagnostics of cardiovascular diseases. Also, generating new synthetic ECGs will allow us to solve the issue of the lack of labeled ECG for use them in supervised learning.